1. Introduction

Optical parametric interactions are ultra-fast in principle, and hence, should offer ultra-broad spectra because they are based on non-resonant electronic response to the applied optical field. However, the practical bandwidth becomes extremely narrow if one implements phasematching or quasi-phase matching (QPM) in order to coherently accumulate the desired harmonic field along the propagation length in the medium. In the time-domain propagation, such bandwidth limit appears as a temporal pulse walk-off between the interacting waves. In ultra-fast pulse applications, thin nonlinear crystals are conventionally used to obtain the required bandwidth because it is inversely proportional to the propagation length in the firstorder approximation. However, it is a poor trade-off between the bandwidth and the efficiency, since the latter is inversely proportional to the square of the length.

An approach to solve this bandwidth problem is the group-velocity matching (GVM) technique. Efficient broadband second-harmonic generation (SHG), for example, has been realized in a periodically poled lithium niobate (PPLN) crystal [1]. The fundamental idea was to match the group-velocities between the fundamental and the second harmonic in the desired spectral regions by using the material dispersion, while high conversion efficiency was achieved by quasi-phase-matching (QPM). The method of utilizing the GVM concept has advantages of its simplicity in the QPM design and fabrication, collinear interaction, and hence better output beam profiles compared to the other broadband methods employing aperiodic QPM gratings [2] or non-collinear beam configuration [3].

The GVM method in QPM SHG can also be applied to broadband QPM optical parametric amplification (OPA), where the QPM period and the pump wavelength are chosen to match the group-velocities of the signal and the idler in the desired spectral region. We recently demonstrated efficient OPA of ~100 nm broad signals in the communication band [4]. Ultrabroadband optical parametric generation (OPG) has been recently demonstrated by pumping around the point of zero-group velocity dispersion in orientation-patterned GaAs [5], periodically poled KTP [6], and PPLN [7]. Pumping around the point of zero-group velocity dispersion is equivalent to the GVM application near the degeneracy point.

In this paper we report an efficient ultra-broad OPG gain band in PPLN using the GVM method near the degeneracy point. Due to the extremely broad OPG band, temperatureinsensitive cascaded visible light generation has been observed. With a proper design of QPM period and pump wavelength, we could also obtain strong blue radiation, resulting in the generation of white light made of all three primary colors with a single pump source.

2. Design consideration for ultra-broadband QPM OPG and RGB generation

Design rules for GVM OPG (OPA) are described elsewhere [4]. The key idea was to find out proper sets of parameters such as QPM period, pump wavelength, and temperature that make the group-velocities of the signal and idler equal by using the natural dispersion of lithium niobate. The OPG efficiency was maximized by the QPM technique. Here, we sought for much broader band than those in ref. 4 by bringing the signal and idler closer to the degeneracy point. There are several sets of parameters (QPM period, pump wavelength and temperature) satisfying ultra-broadband OPG, but they are subject to more restrictions for RGB generation.

As we will verify later, red and green beams are naturally generated by high-order QPM sum-frequency generation (SFG) between the pump and the signal/idler, once we have obtained a broad OPG band. Thus, producing specific color peaks depends on the SFG processes, in which the pump wavelength picks up only quasi-phase-matched ones in the OPG band. For efficient blue generation we selected the pump wavelength close to the high-order QPM SHG in blue, putting an additional restriction on the GVM design. Only a narrow range of parameter space allows both broad OPG and blue SHG. Although temperature can be a free parameter to some extent, the pump wavelength and the QPM period have limited acceptable ranges. Here, we fixed QPM period at 27.0 µm, and tried to tune the sample temperature and the pump wavelength (only slightly).

Figure 1 (solid curve) shows the numerical estimation of OPG band at temperature of 24°C for a pump wavelength of 933 nm. The calculation was based on the dispersion of the extraordinary index of congruent LiNbO3 crystal determined by Jundt [8]. The zero-GVD point (1918 nm) does not necessarily coincide with the degeneracy point (933×2 nm) as was mentioned in ref. 5 and 6. Although this is not an optimal spectrum for OPG, the pump around 933 nm can generate sixth-order QPM SHG in blue, according to the dispersion in ref. 8.

 

Fig. 1. OPG spectrum, Symbols: experiment, line: calculation using Sellmeier’s formula. Pump: 933 nm, for 20 mm-long PPLN with QPM period of 27.0 µm at 24 oC. (Dips at λ₁~λ₄ are explained in the text.)

Download Full Size | PDF

3. Experiment

The pump light source was generated from the OPG-OPA system using two β-BBO crystals, which was pumped by the third-harmonic (355 nm) of a mode-locked Nd:YAG laser (Quantel YG901; pulse width, 35 ps; repetition rate, 10 Hz) [4]. The output wavelength of the OPGOPA was tuned from 910 nm to 950 nm, in which range we could observe an obvious changes in the OPG spectrum from the PPLN crystal. The pump beam was loosely focused into a 20 mm-long PPLN crystal with a QPM period of 27.0 µm. Temperature of the sample could be varied from 10 to 60°C with an accuracy of 0.1°C. A band pass filter which transmits from 1.2 to 2.5 µm was used to cut the pump. The OPG spectrum was measured with a monochromator (SPEX 1702) and a PbS detector, which was corrected with the detector response only. The RGB spectrum was recorded by using a CCD spectrometer.

4. Results and discussion

The OPG spectrum of the PPLN crystal was broadest for the pump wavelength of 933 nm at 24 oC (shown in Fig. 1). The measured OPG band starts from ~1400 nm and ends at ~2600 nm, exhibiting a total bandwidth of about 1200 nm. Because the longer wavelength cut-off was limited by the transmission window of the band-pass filter, we could not measure the features for the long wavelength end of the spectrum. Except that, a rough agreement is observed between the measured OPG spectrum and the calculated one. In order to verify that this is the broadest spectrum, we observed the change in the OPG spectrum by tuning the pump wavelengths and the sample temperature. Figure 2(a) shows the OPG spectra at three different pump wavelengths, λ_p. For λ_p=933 nm, the coherence length curve l_c(λ) makes nearly tangential contact with the QPM period line (horizontal) including the degeneracy point as shown in Fig. 2(b), resulting in an ultra-broad OPG band. For λ_p=946 nm, l_c(λ) shifts upward, making two point intersections with the horizontal QPM line, explaining the narrow OPG signal band in Fig. 2(a). (The idler band was beyond our detection limit, expected at 3.2 µm.) For λ_p=928 nm, however, QPM cannot be achieved because l_c(λ) does not intersect the horizontal QPM line, resulting in a poor signal. For the broadest OPG spectrum, the integrated energy was 1.7 µJ/pulse for the pump energy of 22 µJ/pulse.

 

Fig. 2. OPG spectra for three different pump wavelength at 24°C. Experimental spectra (a) and coherence length curves versus signal or idler wavelength, shown with 27.0 µm-QPM period line (horizontal) (b). Open triangles and dotted line for λ_p=946 nm, open circles and solid line for λ_p=933 nm, closed squares and dashed line for λ_p=928 nm.

Download Full Size | PDF

Once we obtained a broad OPG spectrum with the pump around 933 nm, a beam of white light was observed collinearly with the pump, signal and idler. A spectroscopic analysis revealed that the white light was made of three primary color beams; red, green and blue (RGB) as shown in Fig. 3 with a photograph. The total energy of the RGB was ~0.5 µJ/pulse at the same pump energy as above. Although the wall-plug efficiency from Nd:YAG laser to the OPG output is very small, it could be significantly improved if the PPLN sample is pumped with a Ti:sapphire laser or a solid-state laser with a good beam profile operating at the designed wavelength.

The R and G beams originate from the broad OPG band. A careful look at the OPG spectrum reveals sharp dips at four wavelengths as indicated in Fig. 1. The observed red peak at 648 nm corresponds to the second-order QPM SFG between the dip at 2122 nm (λ₃ in Fig. 1) and the pump at 933 nm. The green peak at 575 nm is due to the third-order QPM SFG between the dip at 1498 nm (λ ₁) and the pump. Thus the dips at λ ₁ and λ ₃ are interpreted as the depletions due to the generation of G and R light, respectively. The other two depletions at λ ₂ and λ ₄ in the OPG band can be easily understood from the fact that they are the conjugate wavelengths of λ ₃ and λ ₁, respectively. On the other hand, the blue peak at 467 nm is the sixth-order QPM SHG of the pump itself. Even-order QPM processes are also possible due to the fact that the average duty-ratio of actual periodic reversal is slightly different from 1:1, and to the duty-cycle fluctuation of the QPM grating inherent to the fabrication process of PPLN [9].

A more concrete proof for the origin of the RGB generation can be provided by calculating the corresponding SFG and SHG QPM bands for the RGB peaks (solid line in Fig. 3) with pump wavelength of 933 nm for the QPM period of 27.0 µm using the Sellmeier’s formula in ref. 8. An excellent agreement between the calculated QPM bands and the measured R and G peaks leads to the conclusion that they indeed originate from the highorder QPM SFG processes as claimed above. It is noticed that the measured B peak (dotted curve in Fig. 3) is stronger than the G and R peaks although it is not exactly quasi-phase matched (according to the given Sellmeier’s formula). The conversion efficiencies of the G and R beams could be further improved by using an amplifier stage (or double-pass in the crystal).

 

Fig. 3. Measured white light spectrum (dotted) and calculated QPM peaks locations (solid) along with a photograph of the spectrally separated white light.

Download Full Size | PDF

Finally, we investigated the temperature dependence of the RGB spectrum because stability against temperature change in such device would be one of the most desired properties in practical applications. Figure 4 shows typical measured RGB spectra at 24°C (solid) and 55°C (dotted), keeping the pump wavelength fixed. It is clear that the changes in the peak positions and intensities are not very sensitive to temperature change. The changes in the spectral locations of the R and G peaks were less than 1 nm for the temperature change from 24 to 55°C. A shift of 1 nm in the R or G peaks corresponds to a shift in a few nm in the IR-OPG spectrum. Although the OPG band changes with temperature, it is still broad enough to produce R and G peaks.

 

Fig. 4. Measured RGB spectra at 24°C (solid) and 55°C (dotted)

Download Full Size | PDF

5. Conclusion

We have designed and experimentally demonstrated an ultra-broad OPG band of 1200 nm, using a single PPLN crystal. It was accomplished by the group-velocity matching between the signal and the idler near the degeneracy point. We also achieved simultaneous RGB generation in a single PPLN crystal at a fixed pump wavelength. The intensities and spectral locations of the primary color peaks were quite robust against temperature variation, which is one of the important requirements for practical applications.